Effect of Prenatal Exposure to Airborne Polycyclic Aromatic Hydrocarbons on Neurodevelopment in the First 3 Years of Life among Inner-City Children

Frederica P. Perera; Virginia Rauh; Robin M. Whyatt; Wei-Yann Tsai; Deliang Tang; Diurka Diaz; Lori Hoepner; Dana Barr; Yi-Hsuan Tu; David Camann; Patrick Kinney


Environ Health Perspect. 2006;114(8):1287-1292. 

In This Article

Materials and Methods

The present cohort study is being conducted by the Columbia Center for Children's Environmental Health (CCCEH) (Perera et al. 2003). The study was approved by the Institutional Review Board (IRB) of Columbia University. Dominican and African-American women (ethnicity classified by self-report) residing in Washington Heights, Central Harlem, and the South Bronx, New York, who registered at the obstetrics/gynecology clinics at New York Presbyterian Medical Center and Harlem Hospital by the 20th week of pregnancy were approached in the clinics for consent. At that time, the women agreeing to participate in the prospective cohort study signed the IRB-approved consent form. Eligible women were nonsmokers during the current pregnancy; were free of diabetes, hypertension, and known HIV; had no documented or reported drug abuse; and had resided in the area for at least 1 year. At the time of this report, of 648 consenting and eligible mother–infant pairs, 536 were still participating in the cohort study; 271 children had reached 3 years of age. The retention rate for the full cohort was 83% at the 3-year follow-up. There were no significant differences between women retained in the study versus those who were lost to follow-up, on maternal age, ethnicity, marital status, education, income, gestational age, or birth weight of the newborn.

In this report we focus on the 183 children 3 years of age who had valid prenatal PAH monitoring data, all three annual developmental assessments, prenatal questionnaire data on ETS, measurements of cotinine in maternal and cord blood samples ≤ 25 ng/mL (to exclude the possibility that the mother was an active smoker), and CPF level in cord blood. This group did not differ in any of the maternal or infant characteristics or prenatal exposures in Table 1 from the 80 children 3 years of age excluded from the analysis because of missing data. Of these,64 children were excluded because of missing developmental testing data.

A 45-min questionnaire was administered by a trained bilingual interviewer during the last trimester of pregnancy (Perera et al. 2003). The questionnaire elicited demographic information, residential, health, and environmental history, including active and passive smoking [household members who smoke and estimated cigarettes smoked per day by smoker(s)], and socioeconomic information related to income and education. Postnatal interviews were administered at 6 months, annually, and every 3–6 months in between to determine any changes in residence, exposure to ETS, and other health or environmental conditions.

During the third trimester of pregnancy, personal monitoring was carried out as previously described (Perera et al. 2003). Vapors and particles ≤ 2.5 µg in diameter were collected on a precleaned quartz microfiber filter and a precleaned polyurethane foam cartridge backup. The samples were analyzed at Southwest Research Institute (San Antonio, TX) for benz[a]anthracene, chrysene, benzo[b]fluroanthene, benzo[k]fluroanthene, BaP, indeno[1,2,3-cd]pyrene, disbenz[a,h]anthracene, and benzo[g,h,i]perylene as described by Tonne et al. (2004). For quality control, each personal monitoring result was assessed as to accuracy in flow rate, time, and completeness of documentation. All of the 183 subjects had samples of acceptable quality.

A sample of umbilical cord blood (30–60 mL) was collected at delivery by syringing blood into a heparinized syringe to avoid clotting. A sample of maternal blood (30–35 mL) was collected within 2 days postpartum into heparinized Vacutainer tubes (BD Medical,Franklin Lakes, NJ) by hospital staff. Samples were processed at the CCCEH laboratory, and portions were sent to the Environmental Health Laboratory at the Centers for Disease Control and Prevention (CDC; Atlanta, GA) for analysis of cotinine, heavy metals, and pesticides. Plasma cotinine was analyzed using high-performance liquid chromatography atmospheric-pressure ionization tandem mass spectrometry as described by Bernert et al. (1997, 2000). Plasma levels of CPF were analyzed using isotope-dilution gas chromatography–high-resolution mass spectrometry as described by Barr et al. (2002). In a subset (n = 135) of subjects, lead was analyzed by inductively coupled plasma mass spectrometry (CDC 2003).

Information was abstracted by the research workers from mothers' and infants' medical records after delivery, including gestational age at birth, infant sex, birth weight, length, head circumference, infant malformations, and pregnancy complications. Gestational age was based on medical records for almost all subjects. Where those data were missing, gestational age was calculated from the last menstrual period.

We used the Bayley Scales of Infant Development–Revised (BSID-II) to assess cognitive and psychomotor development at 12, 24, and 36 months of age (Bayley 1993). The BSID-II is the most widely used norm-referenced developmental test for young children, can be used to diagnose developmental delay, and is known to be sensitive to the developmental effects of toxic exposures such as low-level intrauterine lead. The stability of cognitive assessments during the first few years of life is limited, but the predictive power increases after 2 years. When administered at 3 years of age, the BSID-II has moderate predictive power for subsequent intelligence and school performance and is clinically useful for the identification of children performing in the subnormal range (Bayley 1993; Burchinal et al. 2000; Sternberg et al. 2001). Each test yields a developmental quotient (raw score/chronologic age), which generates a mental development index (MDI) and a corresponding psychomotor development index (PDI). In addition, children are classified as normal (> 85), moderately delayed (> 70 and ≤ 85), or severely delayed (≤ 70) based on standardized cut-points. Each child was tested under controlled conditions at the CCCEH by a bilingual research assistant, trained and checked for reliability. In the present study, the interrater reliability for the 24-month MDI was r = 0.92, based on double scoring of a random 5% of the sample (Rauh et al. 2004). One hundred eighty-one children had complete MDI at 1, 2, and 3 years of age; 181 had complete data on PDI, and 183 had either complete MDI or PDI.

Behavior problems were measured by maternal report on the 99-item Child Behavior Checklist (CBCL) for children 1.5–5 years of age, which collects information on child behaviors occurring in the past 2 months (Achenbach and Rescorla 2000). The CBCL is well validated, easy to administer, and useful as a screen for behavior problems. The Total Problems (T) score is the sum of the scores on the specific problem items plus the highest score on any additional problems entered by the respondent for the open-ended item 100, and is computed by summing the scores for the problems. T scores >63 (> 90th percentile) represent the clinical range, and T scores between 60 and 63 (83rd to 90th percentile) represent the borderline range. The CBCL also yields scales derived from the Diagnostic and Statistical Manual of Mental Disorders (2000) that are intended to approximate clinical diagnoses, including affective, anxiety, pervasive developmental, attention deficit/hyperactivity, and oppositional defiant problems. All subscales are scored continuously and also categorically using a borderline or clinical cut-point corresponding to the 98th percentile for each domain. One hundred sixty-eight children of the children with Bayley scores also had CBCL data.

Maternal nonverbal intelligence was measured by the Test of Non-Verbal Intelligence, second edition (Brown et al. 1990), a 15-min, language-free measure of general intelligence, relatively stable and free of cultural bias. The test was administered when the child was 3 years of age. The quality of the proximal caretaking environment was measured by the Home Observation for Measurement of the Environment (Caldwell and Bradley 1979), administered at 3 years of age. The instrument assesses physical and interactive home characteristics (Bradley et al. 1989), is predictive of developmental scores in early childhood, and has been widely used in studies of neurotoxicity (e.g., Bellinger et al. 1988).

As in prior analysis (Perera et al. 2003), a composite PAH variable was computed from the eight intercorrelated PAH air concentration measures (r values ranging from 0.34 to 0.94; all p-values < 0.001 by Spearman's rank). This variable was dichotomized at the fourth quartile (4.16 ng/m3) to obtain a measure of high/low exposure that is more robust than the continuous variable. The CPF variable was also dichotomized at the fourth quartile as previously described (Whyatt et al. 2004). The concentration of lead in cord blood was treated as a continuous variable.

We estimated the associations between prenatal PAH exposure (high/low) and developmental scores (cognitive and psychomotor) for 12, 24, and 36 months of age using multiple linear regression for continuous outcomes (MDI and PDI) and logistic regression for categorical outcomes (likelihood of being classified as developmentally delayed). We estimated associations between prenatal PAH exposure (high/low) and behavior problems in the clinical range at 36 months of age using logistic regression. We used general estimated equation (GEE) (Liang and Zeger 1986) to estimate the size of the PAH effect over time (through 36 months) and at specific time points. To evaluate trends over time, the model compares the two exposure groups in terms of the difference in MDI scores obtained at 1 year of age (baseline) versus 2 years and 1 year versus 3 years, respectively. GEE has the advantage of requiring fewer assumptions than other methods; thus, the results of GEE are more robust. All effect estimates, 95% confidence intervals (CIs), and p-values (α = 0.05) were generated using SPSS (version 11.5; SPSS Inc., Chicago, IL) and SAS (version 9.0; SAS Institute Inc., Cary, NC). Covariates were retained in the models as potential confounders if they exhibited a relationship (p ≤ 0.1) with motor or mental development, regardless of their association with PAH exposure. The final models included an indicator for PAH exposure, the child's exact age at test administration, child's sex, ethnicity, gestational age at birth, quality of the home (caretaking) environment, and prenatal exposure to ETS and CPF measured as described above. In addition, the possible confounding of prenatal exposure by lead was tested in the subset of 135 children with available data. Interactions of PAH exposure with other independent variables were tested as appropriate.

We assessed potential mediation of the association between PAH exposure and development over time by including those fetal growth parameters previously shown to be affected by prenatal PAH exposure (birth weight and head circumference) in the models. If the estimate of the PAH effect on neurodevelopment was attenuated in the presence of a fetal growth parameter, mediation was considered to be present.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: